📎 Webclip
Object-Relational Mapping is the Vietnam of Computer Science
Atwood accepts Ted Neward’s argument that the object/relational mismatch has no good solution. He says developers keep running into the gap between relational databases and object models, and that every proposed fix adds serious trade-offs that may only become clear much later in the project.
Reading notes#
- He describes ORM as a long-standing mismatch between relational models and traditional object models.
- He says there is no good solution to the problem, only approaches with painful trade-offs.
- He lists several possible responses: abandoning objects, embracing object storage, writing manual mapping code, using ORM only for part of the problem, integrating relational ideas into languages, and building frameworks around relational constructs.
- He notes that raw relational access can bypass ORM caching.
- He concludes that the workable choice is to pick one side, either objects or relational databases, so the mapping problem disappears.
